fix: don't portray special schools as failing the mainstream benchmark

Special schools, PRUs and alternative provision teach pupils with SEND who
sit the same KS2/KS4 assessments but very few reach the mainstream "expected
standard". Their headline attainment is therefore ~0% (or a very low
Attainment 8), and the site was comparing that to the England average and
painting it red — e.g. Greenmead School (a community special school) rendered
as "0.0% — −62 pts below England average" with three 0% red SATs bars. That
portrays a special school as catastrophically failing against a benchmark
that doesn't fit it.

Add a shared `isSpecialSchool()` helper (detects every DfE special-school
establishment type — all contain "special" — plus PRUs / alternative
provision) and drop the mainstream England comparison + "below" framing for
these schools across every surface:

- Detail (primary + secondary): a plain-English context note explaining the
  school is special and why the comparison isn't shown; England-average delta
  chips, "England avg" hints, the SATs national markers, the Attainment-8
  "vs national" bar and the trend chart's England overlay are all suppressed.
  An all-zero placeholder SATs row hides the (empty) subject bar chart and the
  "why is combined lower" bridge.
- Rankings / search rows (primary + secondary): the mainstream RWM / Attainment
  8 stat shows "—" with no "vs national" delta, instead of "0% · −62 vs
  national".
- Compare: special schools' attainment values are dropped (no misleading dot
  at 0% / no "Below England average" chip); progress banding, which IS a fair
  measure for special schools, is kept.

Belt-and-braces zero-guard: a whole-row zero attainment (special or a
suppressed cohort) is also treated as not-comparable, while a legitimate
single 0 (e.g. 0% exceeding at a mainstream school) stays comparable.

Tests: new isSpecialSchool unit tests (every DfE special type matched, no
mainstream false positives); an e2e journey asserts Greenmead shows the
special-school note and no England-average comparison. tsc clean; 108/108 unit.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0146VHeLAWjDVE2B5uU67jCB

nextjs-app/lib/utils.ts

@@ -795,3 +795,24 @@ export const PROPOSED_TO_CLOSE_STATUS = 'Open, but proposed to close';
 export function isProposedToClose(school: { status?: string | null }): boolean {
   return school.status === PROPOSED_TO_CLOSE_STATUS;
 }
+
+/**
+ * Special schools, pupil referral units and alternative provision teach pupils
+ * with SEND or outside mainstream settings. Their pupils sit the same KS2/KS4
+ * assessments but very few reach the mainstream "expected standard", so the
+ * headline attainment measures — and any comparison to the England average —
+ * are not a fair judgement of the school. Callers use this to drop the
+ * mainstream-benchmark framing (deltas, "below England", national markers)
+ * rather than portray these schools as failing.
+ *
+ * Detection is by establishment type: every DfE special-school type contains
+ * "special" (e.g. "Community special school", "Academy special converter/sponsor
+ * led", "Foundation special school", "Non-maintained special school", "Free
+ * schools special", "Other independent special school"); PRUs and alternative
+ * provision are matched by name. Special schools carry a mainstream `phase`
+ * (Primary/Secondary/All-through), so `phase` alone can't identify them.
+ */
+export function isSpecialSchool(school: { school_type?: string | null }): boolean {
+  const t = (school.school_type ?? '').toLowerCase();
+  return /\bspecial\b/.test(t) || /pupil referral/.test(t) || /alternative provision/.test(t);
+}